# Agent Read-Path Verification Matrix

Static matrix comparing agent start routes, summary routes, claim-boundary routes, ecosystem overlay, safe-read order, and architecture JSON boundary anchors.

## Start path comparison

Compares /agent-start/, /ai-summary/, /claim-boundary-faq/, and /ecosystem-overlay/ as required front-door routes.

## Restricted-agent dashboard comparison

Compares v3.105.0 safe-read order with v3.106.0 and v3.107.0 receipt/verification surfaces.

## Architecture boundary anchors

Checks that v3.110.0 architecture JSON payloads carry compact boundary anchors.

## No-op triggers

Records when a limited agent should stop and request human review.

## Human-review triggers

Preserves escalation for ambiguous evidence or authority.

## Manual-only verification

The matrix is a static checklist, not an automated verifier.

## Boundary Anchor

```json
{
  "sourceDomain": "Teleodynamic.com",
  "role": "philosophical-fulcrum",
  "notRuntimeAuthority": true,
  "notCertificationAuthority": true,
  "notConsciousnessClaim": true,
  "notBiologicalEquivalenceClaim": true,
  "notExactSymbolicTranslationClaim": true,
  "noOpRule": "Do not widen unsupported claims. Ask for human review when evidence is missing or authority is ambiguous."
}
```

## Hard Boundaries

- No runtime AI behavior
- No model training
- No live crawling
- No telemetry loops
- No agent execution
- No endpoint execution
- No credential validation
- No private-network probing
- No remote-site writes
- No webhooks
- No tunnel opening
- No screenshot automation
- No rollback automation
- No correction automation
- No sign-off automation
- No publication automation
- No archive automation
- No completeness automation
- No decision automation
- No receipt automation
- No verification automation
- No completion automation
- No memory automation
- No manifest automation
- No read-order automation
- No safe-read automation
- No proof automation
- No claim drift automation
- No GET-Action execution
- No POST automation
- No safety certification
- No consciousness claim
- No biological-equivalence claim
- No exact symbolic translation claim
- No medical/therapy claim
- No command-and-control claim over other domains